Comcast trials NVIDIA GPUs at network edge to support AI services

Comcast Corporation (NASDAQ:CMCSA) said it has begun a field trial deploying NVIDIA GPUs at the edge of its network to evaluate artificial intelligence applications that can run closer to end users. The initiative is intended to lower latency by processing AI workloads in regional network facilities rather than in distant centralized data centers.

The test will use Comcast’s distributed network infrastructure, which reaches about 65 million homes and businesses. The company plans to evaluate three initial applications: a personalized advertising platform built on Decart’s real-time AI video models, an AI-powered small business concierge service using Personal AI technology on HPE ProLiant servers, and lower-latency gaming capabilities.

The personalized advertising system will generate video advertisements tailored at the household level, incorporating factors such as language preferences and demographic information. The small business concierge service is designed to provide AI-based customer greetings, appointment scheduling and operational assistance. Meanwhile, the gaming use case builds on Comcast’s existing low-latency infrastructure that supports NVIDIA GeForce NOW.

“The industry is shifting towards a more distributed AI infrastructure and Comcast operates a network that supports it today,” said Elad Nafshi, Chief Network Officer at Comcast. The company said its network includes DOCSIS 4.0 FDX nodes, smart amplifiers and intelligent gateways designed to enable real-time AI inference.

“Distributed AI Grid is the next big opportunity for the telecommunications industry, and Comcast’s nationwide, deeply distributed network is a perfect match for building it,” said Ronnie Vasishta, SVP of AI and Telecoms at NVIDIA.

Comcast and NVIDIA said they will also explore further use cases, including AI-driven advertising, services for small businesses, premium gaming offerings and edge computing services for third parties. Comcast plans to provide further details about the project during NVIDIA’s GTC conference in San Jose.


Posted

in

by

Tags: